External/wayback-machine-downloader
1
0
Fork 0
mirror of https://github.com/StrawberryMaster/wayback-machine-downloader.git synced 2025-12-29 16:16:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add exact_match option.

Browse Source 
With this option set, Wayback Machine Downloader
will only look for snapshots matching the exact base_url
passed in rather than base_url and its children.

This is useful when trying to download a single file
rather than mirroring a site.
This commit is contained in:
Oleg Pudeyev
2017-03-15 17:58:05 -04:00
parent 65b1948517
commit d926f965f9
1 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
lib/wayback_machine_downloader.rb
View File

@@ -16,11 +16,13 @@ class WaybackMachineDownloader
VERSION = "1.1.5"
attr_accessor :base_url, :directory, :from_timestamp, :to_timestamp,
attr_accessor :base_url, :exact_match, :directory,
:from_timestamp, :to_timestamp,
:only_filter, :exclude_filter, :all, :maximum_pages, :threads_count
def initialize params
@base_url = params[:base_url]
@exact_match = params[:exact_match]
@directory = params[:directory]
@from_timestamp = params[:from_timestamp].to_i
@to_timestamp = params[:to_timestamp].to_i
@@ -84,8 +86,10 @@ class WaybackMachineDownloader
snapshot_list_to_consider = ""
snapshot_list_to_consider += get_raw_list_from_api(@base_url, nil)
print "."
snapshot_list_to_consider += get_raw_list_from_api(@base_url + '/*', nil)
print "."
unless @exact_match
snapshot_list_to_consider += get_raw_list_from_api(@base_url + '/*', nil)
print "."
end
@maximum_pages.times do |page_index|
snapshot_list = get_raw_list_from_api(@base_url + '/*', page_index)
break if snapshot_list.empty?